Closed-loop visual control method and system based on image enhancement and YOLO identification

Abstract

The invention discloses a closed-loop visual control method and system based on image enhancement and YOLO recognition, and relates to the technical field of data processing, and the method comprises the steps: obtaining travel data, carrying out the feature extraction and fusion of original data through an auto-encoder, carrying out the smoothing of the travel data through time series data analysis, and predicting a trend; performing sentiment analysis on the comment data, constructing a sentiment similarity matrix, constructing a sentiment network by using a graph theory technology, and dividing the sentiment network; based on an LSTM deep learning structure, constructing a scenic spot popularity prediction model; optimizing a prediction model based on the division result of the emotion network; performing multi-objective optimization on strategy generation by using an evolutionary algorithm; generating a strategy decision tree according to the emotion network, the prediction model and the optimization result; and generating an optimal travel path according to the current situation of the user and the decision tree dynamic adjustment strategy. The business travel data processing method provided by the invention is more accurate in prediction result, and enhances the user experience and the business travel satisfaction degree.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a closed-loop vision control method and system based on image enhancement and YOLO recognition. Background Art

[0002] As the number of business travelers continues to grow, a vast amount of travel data is generated and stored, including but not limited to basic information about scenic spots, user reviews, historical records of popular visits, user behavior data, and user contextual data. Traditional methods for processing travel data rely primarily on simple statistical analysis or basic time series forecasting techniques, lacking a deep understanding of complex emotions and user behavior. These methods suffer from significant deficiencies in prediction accuracy, data fusion capabilities, and personalized recommendations. This results in suboptimal travel recommendations and a limited user experience.

[0003] To address the shortcomings of traditional methods, this technology proposes a novel approach to travel data processing. Combining deep learning, time series analysis, sentiment analysis, and graph theory, this method enables in-depth mining and analysis of travel data from multiple dimensions. Specifically, it can more accurately predict popular tourist attractions, better understand user sentiment through sentiment analysis, and provide more precise and personalized travel route recommendations. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: the existing traditional travel data processing methods have the problems of low accuracy, inability to deeply understand user emotions and lack of personalized recommendations, as well as how to effectively integrate multiple technical means, improve prediction accuracy and achieve accurate travel route recommendations.

[0006] To solve the above technical problems, the present invention provides the following technical solution: a closed-loop visual control method based on image enhancement and YOLO recognition, comprising:

[0007] Acquire travel data, use autoencoders to extract and fuse features from the raw data, and then use time series data analysis to smooth the travel data and predict trends. Perform sentiment analysis on the review data and construct a sentiment similarity matrix. Use graph theory to build and segment the sentiment network. Build a model to predict attraction popularity based on the LSTM deep learning architecture. Optimize the prediction model based on the segmentation results of the sentiment network. Use evolutionary algorithms to perform multi-objective optimization on strategy generation.

[0008] Generate a strategic decision tree based on the emotional network, prediction model, and optimization results; dynamically adjust the strategy based on the user's current situation and decision tree to generate the optimal travel path.

[0009] As a preferred solution of the travel data processing method described in the present invention, the acquired travel data includes basic information of scenic spots, user evaluation data, historical records of classic visits, user behavior data, and user current scenario data; the feature extraction includes using an autoencoder to convert the travel data into a feature vector V; the fusion includes weighted combination of the feature vectors extracted by the autoencoder and normalization processing to form a unified feature representation.

[0010] As a preferred embodiment of the travel data processing method of the present invention, the trend prediction includes decomposing the smoothed data into trend, season, and residual using STL, fitting the trend component in the STL decomposition using an ARIMA model, determining parameters and establishing a model, and predicting the trend for a preset time using the ARIMA model;

[0011] The decomposition of trend, season and residual is expressed as,

[0012] Y t =T t +S t +R t

[0013] Among them, Y t represents the smoothed travel data at time point t, T t represents the trend component at a given time point, S t represents the seasonal component of the time point, R t represents the residual component at time point t;

[0014] The model is expressed as:

[0015] (Y t -αV t ·β)=φ(B)·(1-B) d ·(Y t-1 -αV t-1 ·β)+θ(B)·R t

[0016] Among them, α represents the weight parameter, V t represents the eigenvector at time point t, β represents the weight of the eigenvector V, φ(B) represents the autoregressive parameter, θ(B) represents the moving average parameter, B represents the shift operation, and d represents the number of differences;

[0017] The prediction includes if S t If there is a periodic pattern within the preset time, determine whether the seasonal component persists and obtain S tIn the future forecast value, if the forecast shows that the seasonal component maintains periodicity in the future, it is judged to be a cyclical trend. If the forecast shows that the periodicity of the seasonal component does not maintain periodicity, it is judged that the future T t Value, if T t The growth rate is greater than the preset threshold, and it is judged to be in an upward trend. If the T t The growth rate is less than the preset threshold, and it is judged to be in a downward trend. If the T t If the growth rate is equal to the preset threshold, it is judged to be in a stable trend.

[0018] As a preferred solution of the travel data processing method of the present invention, the sentiment analysis is expressed as:

[0019] S i =f(C i )

[0020] S′ i =S i ×W T

[0021] Among them, f represents the sentiment analysis function, C i represents the i-th user comment, S i represents the sentiment score of the i-th user comment, S′ i Indicates C i The sentiment score, W T Indicates the trend weight of travel data;

[0022] When S′ i When S′ i = 0, neutral sentiment, when S′ i When it is >0, it is judged as positive emotion;

[0023] The constructed sentiment similarity matrix is ​​expressed as:

[0024] M ij =V(C i )·V(C j )×S′ i ×S′ j

[0025] Among them, V(C i ) represents the word embedding vector of the i-th comment, M ij represents the sentiment similarity matrix, C j represents the jth user comment, M ij Comment C i and C j The emotional similarity between j represents the sentiment score of the jth comment;

[0026] The division of the emotional network is expressed as:

[0027] W ij =M ij ×(1+δ|S′ iprev -S′ i |)

[0028] Among them, W ij Comment C i and C j The adjusted sentiment similarity weight, δ represents the weight adjustment parameter, Indicates that C was commented on in the previous cycle i Adjusted sentiment scores;

[0029] The emotional network is divided into two parts: i and S′ j When the sentiment scores are the same, the sentiment similarity weight W is judged ij Combined with the topic classification in the comments.

[0030] As a preferred solution of the travel data processing method of the present invention, the construction of the scenic spot popularity prediction model is expressed as:

[0031] P t+1 =LSTM(S′ t ,H t )

[0032] Among them, P t+1 represents the predicted popularity value of the scenic spot at time point t+1, H t represents the LSTM hidden state at time point t;

[0033] The optimization prediction model is expressed as:

[0034]

[0035] in, represents the real popularity of scenic spots, MSE represents the mean square error loss function, λ represents the regularization coefficient, S′ i,t represents the sentiment score of the i-th user comment at time t, S′ j,t represents the sentiment score of the jth user comment at time t.

[0036] As a preferred solution of the travel data processing method of the present invention, the multi-objective optimization includes predicting the popularity of the scenic spot based on the prediction model of the scenic spot popularity. t+1 , get the actual popularity of attractions Calculate the difference measure δ between the predicted and actual popularity t, if δ t If δ is greater than the preset threshold, it enters the decision adjustment stage. t is less than or equal to a preset threshold, maintaining the optimization prediction model adjustment strategy;

[0037] The decision adjustment stage includes the following steps: t , adjust the emotional network division strategy, calculate the decision adjustment value Θ(P) of the strategy, if Θ(P)>0, highlight the special activities and cultural displays of the scenic spots, provide users with interactive games related to the scenic spots, provide coupons and discount information related to the scenic spots, skip the peak hours, use navigation to avoid congested areas and recommend tour routes, if Θ(P)<0, consider recommending attractions with slightly lower scores but unique cultural or historical backgrounds, provide users with a more high-end or unique travel experience, increase the tour time, and provide interpretations of local culture and customs;

[0038] The adjustment of the emotion network division strategy includes calculating the emotion score change and adjusting the emotion score threshold according to the change;

[0039] The emotion score change is expressed as,

[0040] ΔS=S′-S′ p

[0041] Among them, S′ represents the adjusted sentiment score of the current comment, S′ p represents the adjusted sentiment score of the previous comment, and ΔS represents the difference in sentiment score between the current comment and the previous comment;

[0042] The adjustment sentiment score threshold is expressed as:

[0043]

[0044] Where σ represents the positive sentiment threshold adjusted according to the change in sentiment score, T n represents the new positive threshold, T o represents the old positive threshold;

[0045] Re-evaluate sentiment based on the adjusted sentiment score threshold, using T n Compare with S', if S'>T n , update the label to positive sentiment, if S′<T n , update the label to negative sentiment, if a comment has S′=T n , update the label to neutral sentiment, record the original sentiment label and the adjusted sentiment label of the comment, and continuously monitor the sentiment score;

[0046] The decision adjustment value is expressed as,

[0047] Θ=α×f1+β×f2+γ×f3-λ×δ

[0048] Where P′ represents the optimized predicted popularity of scenic spots, P represents the original predicted popularity of scenic spots, δ represents the difference between the optimized and original predictions, Θ represents the decision adjustment value, f1, f2, and f3 represent the objective functions of satisfaction, cost, and tour time, respectively, and α, β, and γ represent the weight values ​​assigned by users.

[0049] As a preferred embodiment of the travel data processing method of the present invention, the generating of a strategy decision tree includes analyzing user sentiment preference data based on a user sentiment network, using a prediction model to predict future changes in user sentiment, combining the prediction data with the sentiment preference data to identify potential travel strategies, and constructing nodes and branches of the decision tree according to preset weights;

[0050] The dynamic adjustment of the strategy based on the user's current situation and the decision tree includes collecting real-time emotional feedback from the user during the travel process, calculating the deviation between the real-time emotional feedback and the predicted satisfaction in the original prediction model, locating the relevant scenic spot node in the decision tree when the deviation exceeds a predetermined threshold, adjusting the priority of the scenic spot in the strategy of the relevant node, and updating the travel strategy based on the adjusted decision tree;

[0051] Generating the optimal travel route includes listing travel route combinations based on the adjusted decision tree, comprehensively evaluating the emotional satisfaction, travel cost, and time consumption of each route, screening out qualified candidate routes from the evaluated routes based on the user's current location, time, and budget constraints, selecting the route with the highest score from the screened candidate routes for detailed optimization, and determining the optimized route as the optimal route;

[0052] The detailed optimization includes identifying attractions and restaurants that users are interested in and increasing their weight in route decisions, adjusting the user's itinerary based on the estimated arrival time and opening hours of attractions or restaurants, arranging rest periods or relaxing activities between physically demanding activities, optimizing transportation routes to ensure convenient and fast movement between attractions that users are interested in, adjusting restaurant or activity selection based on users' specific food or cultural activity preferences, preparing alternative plans for emergencies such as weather changes, and providing users with a mechanism to fine-tune points of interest during the journey.

[0053] Another object of the present invention is to provide a travel data processing system that can solve the problems of traditional methods in accuracy, sentiment understanding and personalized recommendations by integrating multiple advanced technologies.

[0054] To solve the above technical problems, the present invention provides the following technical solutions: a travel data processing system comprising: a data acquisition module, a feature extraction module, a time series analysis module, a sentiment analysis module, a scenic spot popularity prediction module, a strategy decision tree module, and a route recommendation module; the data acquisition module is configured to collect basic information about scenic spots, user evaluation data, historical records of classic visits, user behavior data, and user current scenario data; the feature extraction module is configured to process the raw data obtained from the data acquisition module and perform feature extraction using an autoencoder; the time series analysis module is configured to analyze time-related travel data, perform trend, seasonal, and residual decomposition on the smoothed data using STL, and predict future trends using an ARIMA model; the sentiment analysis module is configured to analyze user comments, identify users' sentimental tendencies toward scenic spots, and provide sentiment scores for subsequent modules; the scenic spot popularity prediction module is configured to predict the popularity of each scenic spot based on an LSTM deep learning structure and provide prediction results for the strategy decision tree module; the strategy decision tree module is configured to construct a strategy decision tree using the outputs of the previous modules and dynamically adjust the strategy; and the route recommendation module is configured to recommend the optimal travel route for the user by combining the sentiment analysis results, popularity prediction results, and other relevant information.

[0055] A computer device includes a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the travel data processing method described above are implemented.

[0056] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the travel data processing method described above.

[0057] The beneficial effects of this invention are as follows: The travel data processing method provided by this invention combines autoencoders, LSTM, and ARIMA techniques to achieve more accurate prediction results, facilitating scenic spot management and resource allocation. The sentiment analysis module helps businesses capture users' genuine feedback and feelings about scenic spots, providing a basis for optimizing services and improving user satisfaction. Based on users' real-time context, historical behavior, and emotional tendencies, it recommends the most suitable travel routes, thereby enhancing user experience and travel satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort. Among them:

[0059] Figure 1 This is an overall flow chart of a closed-loop visual control method based on image enhancement and YOLO recognition provided by one embodiment of the present invention.

[0060] Figure 2 This is an overall structural diagram of a travel data processing system provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0061] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0062] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0063] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0064] The present invention is described in detail with reference to schematic diagrams. For ease of illustration, cross-sectional views of device structures may be partially enlarged and not to scale when describing embodiments of the present invention. Furthermore, the schematic diagrams are merely illustrative and should not limit the scope of the present invention. Furthermore, in actual production, the three-dimensional dimensions of length, width, and depth should be included.

[0065] In the description of the present invention, it should be noted that the terms "upper, lower, inner, and outer" and other references to orientations or positional relationships are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0066] In this disclosure, unless otherwise specified or limited, the terms "mounted," "connected," and "connected" should be interpreted broadly. For example, they may refer to fixed, removable, or integral connections. They may also refer to mechanical, electrical, or direct connections, indirect connections through an intermediary, or internal communication between two components. Those skilled in the art will understand the specific meanings of these terms in this disclosure.

[0067] Example 1

[0068] Reference Figure 1 , as one embodiment of the present invention, provides a closed-loop visual control method based on image enhancement and YOLO recognition, comprising:

[0069] Obtain travel data, use autoencoders to extract and fuse features from the raw data, and use time series data analysis to smooth the travel data and predict trends.

[0070] Perform sentiment analysis on the comment data and construct a sentiment similarity matrix. Use graph theory technology to build a sentiment network and divide the sentiment network.

[0071] Based on the LSTM deep learning structure, a model for predicting the popularity of tourist attractions is constructed; based on the results of dividing the emotional network, the prediction model is optimized.

[0072] Multi-objective optimization for policy generation using evolutionary algorithms.

[0073] Generate a strategy decision tree based on the sentiment network, prediction model and optimization results.

[0074] Dynamically adjust strategies based on the user's current situation and decision tree to generate the optimal travel path.

[0075] The travel data obtained includes basic information of scenic spots, user evaluation data, historical records of classic visits, user behavior data and user current scenario data; feature extraction includes using autoencoders to convert travel data into feature vectors V; fusion includes weighted combination of feature vectors extracted by autoencoders and normalization processing to form a unified feature representation.

[0076] Trend prediction includes using STL to decompose the smoothed data into trend, season and residual, fitting the trend component in the STL decomposition with the ARIMA model, determining the parameters and establishing the model, and using the ARIMA model to predict the trend of the preset time.

[0077] The decomposition of trend, season and residual is expressed as,

[0078] Y t =T t +S t+R t

[0079] Among them, Y t represents the smoothed travel data at time point t, T t represents the trend component at a given time point, S t represents the seasonal component of the time point, R t represents the residual component at time point t;

[0080] The established model is expressed as,

[0081] (Y t -αV t ·β)=φ(B)·(1-B) d ·(Y t-1 -αV t-1 ·β)+θ(B)·R t

[0082] Among them, α represents the weight parameter, V t represents the eigenvector at time point t, β represents the weight of the eigenvector V, φ(B) represents the autoregressive parameter, θ(B) represents the moving average parameter, B represents the shift operation, and d represents the number of differences.

[0083] Making predictions includes if S t If there is a periodic pattern within the preset time, determine whether the seasonal component persists and obtain S t In the future forecast value, if the forecast shows that the seasonal component maintains periodicity in the future, it is judged to be a cyclical trend. If the forecast shows that the periodicity of the seasonal component does not maintain periodicity, it is judged that the future T t Value, if T t The growth rate is greater than the preset threshold, and it is judged to be in an upward trend. If the T t The growth rate is less than the preset threshold, and it is judged to be in a downward trend. If the T t If the growth rate is equal to the preset threshold, it is judged to be in a stable trend.

[0084] Sentiment analysis is expressed as,

[0085] S i =f(C i )

[0086] S′ i =S i ×W T

[0087]

[0088] Among them, f represents the sentiment analysis function, C i represents the i-th user comment, Si represents the sentiment score of the i-th user comment, S′ i Indicates C i The sentiment score, W T represents the trend weight of travel data, W T ∈{-1,0,1,P}, -1 indicates a downward trend, 0 indicates a stable trend, 1 indicates an upward trend, and P indicates a cyclical trend;

[0089] When in a downtrend, W T The threshold is 1.2, when in a stable trend, W T The threshold is 1, when it is in an upward trend, W T The threshold is 0.8, when in a periodic trend, W T The threshold is 1.1.

[0090] When S′ i When S′ i = 0, neutral sentiment, when S′ i When it is >0, it is judged as positive emotion;

[0091] The sentiment similarity matrix is ​​constructed as follows:

[0092] M ij =V(C i )·V(C j )×S′ i ×S′ j

[0093] Among them, V(C i ) represents the word embedding vector of the i-th comment, M ij represents the sentiment similarity matrix, C j represents the jth user comment, M ij Comment C i and C j The emotional similarity between j represents the sentiment score of the jth comment;

[0094] The emotional network is divided into two parts:

[0095] W ij =M ij ×(1+δ|S′ iprev -S′ i |)

[0096] Among them, W ij Comment C i and C j The adjusted sentiment similarity weight, δ represents the weight adjustment parameter, Indicates that C was commented on in the previous cyclei Adjusted sentiment scores;

[0097] The emotional network is divided into two parts: i and S′ j When the sentiment scores are the same, the sentiment similarity weight W is judged ij Combined with the topic classification in the comments.

[0098] The model for predicting the popularity of scenic spots is constructed as follows:

[0099] P t+1 =LSTM(S′ t ,H t )

[0100] Among them, P t+1 represents the predicted popularity value of the scenic spot at time point t+1, H t represents the LSTM hidden state at time point t;

[0101] The optimized prediction model is expressed as,

[0102]

[0103] in, represents the real popularity of scenic spots, MSE represents the mean square error loss function, λ represents the regularization coefficient, S′ i,t represents the sentiment score of the i-th user comment at time t, S′ j,t represents the sentiment score of the jth user comment at time t.

[0104] Multi-objective optimization includes predicting the popularity of scenic spots based on the popularity model, predicting the popularity of scenic spots P t+1 , get the actual popularity of attractions Calculate the difference measure δ between the predicted and actual popularity t , if δ t If δ is greater than the preset threshold, it enters the decision adjustment stage. t If the value is less than or equal to the preset threshold, the optimization forecast model adjustment strategy is maintained;

[0105] The decision adjustment stage includes the difference metric δ t , adjust the emotional network division strategy, calculate the decision adjustment value Θ(P) of the strategy, if Θ(P)>0, highlight the special activities and cultural displays of the scenic spots, provide users with interactive games related to the scenic spots, provide coupons and discount information related to the scenic spots, skip the peak hours, use navigation to avoid congested areas and recommend tour routes, if Θ(P)<0, consider recommending attractions with slightly lower scores but unique cultural or historical backgrounds, provide users with a more high-end or unique travel experience, increase the tour time, and provide interpretations of local culture and customs;

[0106] Adjusting the sentiment network segmentation strategy includes calculating the sentiment score change and adjusting the sentiment score threshold according to the change;

[0107] The change in sentiment score is expressed as,

[0108] ΔS=S′-S′ p

[0109] Among them, S′ represents the adjusted sentiment score of the current comment, S′ p represents the adjusted sentiment score of the previous comment, and ΔS represents the difference in sentiment score between the current comment and the previous comment;

[0110] Adjusting the sentiment score threshold is expressed as,

[0111]

[0112] Where σ represents the positive sentiment threshold adjusted according to the change in sentiment score, T n represents the new positive threshold, T o represents the old positive threshold;

[0113] Re-evaluate sentiment based on adjusting the sentiment score threshold, using T n Compare with S', if S'>T n , update the label to positive sentiment, if S′<T n , update the label to negative sentiment, if a comment has S′=T n , update the label to neutral sentiment, record the original sentiment label and the adjusted sentiment label of the comment, and continuously monitor the sentiment score;

[0114] The decision adjustment value is expressed as,

[0115] Θ=α×f1+β×f2+γ×f3-λ×δ

[0116] Where P′ represents the optimized predicted popularity of scenic spots, P represents the original predicted popularity of scenic spots, δ represents the difference between the optimized and original predictions, Θ represents the decision adjustment value, f1, f2, and f3 represent the objective functions of satisfaction, cost, and tour time, respectively, and α, β, and γ represent the weight values ​​assigned by users.

[0117] Generating a strategy decision tree involves analyzing user sentiment preference data based on the user sentiment network, using a prediction model to predict future changes in user sentiment, combining the predicted data with sentiment preference data to identify potential travel strategies, and constructing nodes and branches of the decision tree according to preset weights.

[0118] Dynamically adjust the strategy based on the user's current context and decision tree. This includes collecting real-time emotional feedback from users during their travels, calculating the deviation between the real-time emotional feedback and the predicted satisfaction level in the original prediction model, and locating the relevant scenic spot nodes in the decision tree when the deviation exceeds a predetermined threshold. The strategy for the relevant nodes adjusts the priority of the scenic spots, and the travel strategy is updated based on the adjusted decision tree.

[0119] Generating the optimal travel route involves listing travel route combinations based on the adjusted decision tree, comprehensively evaluating each route's emotional satisfaction, travel cost, and time consumption, and then filtering out qualified candidate routes from the evaluated routes based on the user's current location, time, and budget constraints. From these candidate routes, the highest-scoring route is selected for detailed optimization, ultimately determining the optimized route as the optimal route.

[0120] The user's current context includes the user's current time information, location information, and interest information.

[0121] Detailed optimization includes identifying attractions and restaurants that users are interested in and increasing their weight in route decisions, adjusting users' itineraries based on estimated arrival times and opening hours of attractions or restaurants, arranging rest periods or relaxing activities between physically demanding activities, optimizing transportation routes to ensure convenient and fast movement between attractions that users are interested in, adjusting restaurant or activity selection based on users' specific food or cultural activity preferences, preparing alternative plans for emergencies such as weather changes, and providing users with mechanisms to fine-tune points of interest during their journey.

[0122] Using reinforcement learning methods, the path planning system is allowed to self-adjust and learn based on real-time user feedback. Online learning technology is used to continuously update the model to ensure that the system can handle newly emerging data and situations. By comparing user feedback with model prediction results, the model is periodically fine-tuned and overhauled, with fine-tuning performed weekly and overhauled every quarter.

[0123] Use Monte Carlo methods or other simulation techniques to validate strategies against possible future scenarios. If a strategy fails to achieve the expected results in multiple simulations, re-optimize the forecasting model.

[0124] The meta-learning method is used to enable the system to quickly adapt and learn on different travel datasets. When the system faces new and unseen data distributions, it can quickly adjust its strategy based on previous experience.

[0125] Example 2

[0126] Reference Figure 2, which is an embodiment of the present invention, provides a travel data processing system, including: a data acquisition module, a feature extraction module, a time series analysis module, a sentiment analysis module, a scenic spot popularity prediction module, a strategy decision tree module and a route recommendation module; the data acquisition module is used to collect basic information about scenic spots, user evaluation data, historical records of classic visits, user behavior data and user current scenario data.

[0127] The feature extraction module is used to process the raw data obtained from the data acquisition module and perform feature extraction using the autoencoder.

[0128] The time series analysis module is used to analyze time-related travel data, decompose the trend, season and residual of the smoothed data using STL, and predict future trends using the ARIMA model.

[0129] The sentiment analysis module is used to analyze user comments, identify users' emotional tendencies towards attractions, and provide sentiment scores for subsequent modules.

[0130] The attraction popularity prediction module is used to predict the popularity of each attraction based on the LSTM deep learning structure and provide prediction results for the strategy decision tree module.

[0131] The strategy decision tree module is used to use the output of the previous module to build a strategy decision tree and dynamically adjust the strategy.

[0132] The route recommendation module is used to recommend the optimal travel route for users by combining sentiment analysis results, popularity prediction results and other relevant information.

[0133] Example 3

[0134] An embodiment of the present invention is different from the previous two embodiments in that:

[0135] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0136] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0137] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0138] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0139] Example 4

[0140] As an embodiment of the present invention, a closed-loop visual control method based on image enhancement and YOLO recognition is provided. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0141] Experimental environment: Construct a dataset: 6 months of data was captured from three popular travel destinations, including basic information about the destinations, user evaluation data, historical records of classic visits, user behavior data, and user current context data.

[0142] Tools: Python, TensorFlow (for LSTM and autoencoders), statsmodels (for ARIMA models)

[0143] Experimental steps: The above dataset is divided into a training set (70%) and a test set (30%).

[0144] The proposed method and the traditional method are used for training respectively.

[0145] Make predictions on the test set and evaluate the prediction results.

[0146] Analyze user review data and compare the two methods' ability to recognize emotions.

[0147] Travel route recommendations are made to users in the test set, and a user satisfaction survey is conducted. The experimental results are shown in Table 1.

[0148] Table 1 Comparison of experimental results

[0149]

[0150] Our method demonstrates significant advantages in predicting tourist attraction popularity, achieving an accuracy rate of 93%, exceeding the 85% of traditional methods. By combining autoencoders, LSTM, and ARIMA techniques, it effectively captures both long- and short-term dependencies and nonlinear trends in time series. Using STL to decompose the smoothed data into trend, seasonality, and residuals increases the robustness of the predictions, enabling the model to perform well across a wide range of tourist attraction data.

[0151] The proposed method achieved 91% accuracy in sentiment analysis of user reviews, surpassing the 83% achieved by traditional methods. This performance is significantly improved due to the constructed sentiment similarity matrix and sentiment network, which enable the model to more accurately identify and classify user sentiment, particularly in interpreting ambiguous or neutral reviews. This improved sentiment analysis means that tourist attractions can more quickly and accurately capture visitor complaints and suggestions, enabling timely adjustments to further optimize the visitor experience.

[0152] In terms of user satisfaction, our method's travel route recommendations received an 88% satisfaction rating, significantly higher than the 75% achieved by traditional methods. This demonstrates that our method can better integrate users' real-time context, historical behavior, and emotional tendencies to provide more personalized travel routes. Further analysis of the recommended routes reveals that our method often avoids popular attractions during peak hours and recommends less crowded destinations that users may enjoy, enhancing their travel experience.

[0153] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A closed-loop visual control method based on image enhancement and YOLO recognition, characterized in that: include: Obtain travel data, use autoencoders to extract and fuse features from the raw data, and then perform time series data analysis to smooth the travel data and predict trends. Perform sentiment analysis on comment data and construct a sentiment similarity matrix. Use graph theory to construct a sentiment network and divide the sentiment network. Based on the LSTM deep learning structure, a model for predicting the popularity of tourist attractions is constructed; Optimizing the prediction model based on the result of dividing the emotional network; Multi-objective optimization of strategy generation using evolutionary algorithms; Generate a strategy decision tree based on the sentiment network, prediction model and optimization results; Dynamically adjust strategies based on the user's current situation and decision tree to generate the optimal travel path; The trend prediction includes decomposing the smoothed data into trend, season and residual using STL, fitting the trend component in the STL decomposition using an ARIMA model, determining parameters and establishing a model, and predicting the trend of a preset time using the ARIMA model; The decomposition of trend, season and residual is expressed as, Y t =T t +S t +R t Among them, Y t represents the smoothed travel data at time point t, T t represents the trend component at a given time point, S t represents the seasonal component of the time point, R t represents the residual component at time point t; The model is expressed as: (Y t -αV t ·b)=φ(B)·(1-B) d ·(Y t-1 -αV t-1 ·b)+θ(B)·R t Among them, α represents the weight parameter, V t represents the eigenvector at time point t, β represents the weight of the eigenvector V, φ(B) represents the autoregressive parameter, θ(B) represents the moving average parameter, B represents the shift operation, and d represents the number of differences; The prediction includes if S t If there is a periodic pattern within the preset time, determine whether the seasonal component persists and obtain S t In the future forecast value, if the forecast shows that the seasonal component maintains periodicity in the future, it is judged to be a cyclical trend. If the forecast shows that the periodicity of the seasonal component does not maintain periodicity, it is judged that the future T t Value, if T t The growth rate is greater than the preset threshold, and it is judged to be in an upward trend. If the T t The growth rate is less than the preset threshold, and it is judged to be in a downward trend. If the T t If the growth rate is equal to the preset threshold, it is judged to be in a stable trend.

2. The travel data processing method according to claim 1, wherein: The acquired travel data includes basic information of scenic spots, user evaluation data, historical records of classic visits, user behavior data and user current scenario data; The feature extraction includes converting the travel data into a feature vector V using an autoencoder; The fusion includes weighted combination of feature vectors extracted by the autoencoder and normalization to form a unified feature representation. S′>T n , update the label to positive sentiment, if S′<T n , update the label to negative sentiment, if a comment has S′=T n , update the label to neutral sentiment, record the original sentiment label and the adjusted sentiment label of the comment, and continuously monitor the sentiment score; The decision adjustment value is expressed as, Θ=α×f1+β×f2+γ×f3-λ×δ Where P′ represents the optimized predicted popularity of scenic spots, P represents the original predicted popularity of scenic spots, δ represents the difference between the optimized and original predictions, Θ represents the decision adjustment value, f1, f2, and f3 represent the objective functions of satisfaction, cost, and tour time, respectively, and α, β, and γ represent the weight values ​​assigned by users.

Images